Key points are not available for this paper at this time.
We describe a one-pass compression scheme which presumes no statistical properties of the data being compressed. The model structure adaptively selects a subset of first-order Markov contexts, based on an estimate of the candidate context's popularity. The probability distributions for the unselected (lumped) first-order contexts are made the same, reducing cost over a full first-order Markov model. Symbol repetitions are handled in special secondorder Markov contexts. The statistics for each symbol are adaptively determined by an extension of earlier work.
Langdon et al. (Tue,) studied this question.